Industry professionals predict that the price of HBM4 may double next year, driven by the AI trend.

According to a report released by DigiTimes on Friday, industry sources pointed out that due to the dual drivers of the surging demand for artificial intelligence and the structural bottleneck of production capacity, the price of High Bandwidth Memory (HBM) is expected to double by 2027. Industry sources stated that the price of the next generation HBM4 may skyrocket from around $2 per gigabit in the second half of 2026 to $4 to $5 or even higher. On one hand, this is due to the extreme complexity of the manufacturing process of HBM4: its production cycle is up to four to six months long, and the initial yield is significantly low. On the other hand, the production of HBM consumes about three times the wafer capacity of standard DDR5 DRAM, severely limiting the total memory capacity that manufacturers can produce in existing facilities. Further exacerbating the supply shortage is the fact that the three major global HBM producers - Samsung Electronics, SK Hynix, and Micron Technology, Inc. - are signing long-term agreements with top AI customers for three to five years to lock in global memory supply. DigiTimes predicts that by 2027, about half of the total global DRAM production capacity will be completely unavailable to small buyers. While NVIDIA Corporation's upcoming Rubin architecture is accelerating the development process of HBM4, memory manufacturers are also facing an unexpected economic dilemma: the strong performance of the standard server memory market. This year, the profit margins of some suppliers of DDR5 have exceeded 80%, forcing chip manufacturers to demand higher HBM pricing to justify the transfer from traditional DRAM production lines. On Wall Street, this structural tight supply narrative is expected to continue to be a strong catalyst for key storage chip stocks. This Friday, SK Hynix made its debut on the US stock market through American Depositary Receipts (ADR) with a historic $26.5 billion issuance, with a significant premium over the Seoul local stock price, demonstrating a strong trend of massive trading in the artificial intelligence storage sector. Although there are concerns in the market recently about tech giants potentially cutting back on infrastructure spending, according to supply chain channel sources, there will still be fundamental supply shortages for AI hardware by 2027. Therefore, it is expected that by the end of 2026, chip suppliers will maintain an absolute price advantage in contract negotiations, putting consumer electronics manufacturers who have not signed contracts at serious risk of supply shortages. This article is reproduced from Caishang Society, GMTEight Editor: Lifo.